Answers are grounded in this paper and cite their sources.
Machine learning algorithms have been employed to simplify multidimensional self-report scales for efficiently assessing psychological state. This study first conducted a comprehensive evaluation of six feature selection methods to identify optimal strategies for psychological scale reduction and proposed a novel variation of Cronbach’s alpha for shortened scales. This comparison used 19,118 questionnaire responses from offenders in western China. Greedy selection (GREEDY) was first incorporated into the comparison and provided greater computational efficiency and competitive mean absolute error (MAE) in high-item dimensions. Meanwhile, COMBS yielded the lowest mean absolute error (MAE) in low-item dimensions. Moreover, the variation of Cronbach’s alpha ( α ₂ ) outperformed the standard Cronbach’s alpha ( α ₁ ) in 72.7% of dimensions. Using the optimal models, scale length was reduced by 47% on average. In conclusion, we provided a machine learning framework and a novel variation of Cronbach’s alpha for developing offender-specific short forms to achieve efficient psychological screening.
No takes yet. Share an insight, caveat, or question.
Yao et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: